Insights into the Oxidative Stress Response of Salmonella enterica serovar Enteritidis Revealed by the Next Generation Sequencing Approach

As a facultative intracellular pathogen, Salmonella Enteritidis must develop an effective oxidative stress response to survive exposure to reactive oxygen species within the host. To study this defense mechanism, we carried out a series of oxidative stress assays in parallel with a comparative transcriptome analyses using a next generation sequencing approach. It was shown that the expression of 45% of the genome was significantly altered upon exposure to H2O2. Quantitatively the most significant (≥100 fold) gene expression alterations were observed among genes encoding the sulfur utilization factor of Fe-S cluster formation and iron homeostasis. Our data point out the multifaceted nature of the oxidative stress response. It includes not only numerous mechanisms of DNA and protein repair and redox homeostasis, but also the key genes associated with osmotic stress, multidrug efflux, stringent stress, decrease influx of small molecules, manganese and phosphate starvation stress responses. Importantly, this study revealed that oxidatively stressed S. Enteritidis cells simultaneously repressed key motility encoding genes and induced a wide range of adhesin- and salmonellae-essential virulence-encoding genes, that are critical for the biofilm formation and intracellular survival, respectively. This finding indicates a potential intrinsic link between oxidative stress and pathogenicity in non-typhoidal Salmonella that needs to be empirically evaluated.


Introduction
Non-typhoidal Salmonella is a leading cause of foodborne gastroenteritis on the global scale. On the global scale, this foodborne pathogen is responsible for 80 million cases of gastroenteritis annually [1]. The situation in the USA is very similar to the global epidemiological picture of salmonellosis caused by non-typhoidal Salmonella. It has been estimated that from 2000 to 2008, non-typhoidal Salmonella serovars accounted for 1.2 million laboratory-confirmed illnesses, 19,000 hospitalizations, and 380 deaths each year in the USA [2]. The high incidence rate is made more significant by the fact that there have been no signs of decline in the incidence of salmonellosis over the past 15 years in the US. Intriguingly, infections caused by other five major food-borne pathogens (i.e., Escherichia coli O157:H7, Campylobacter spp., Listeria monocytogenes, Yersinia enterocolitica, and Shigella spp.) declined substantially over the same time [2]. Table 1. Primers used in this study for the oxidative stress response gene expression assay and validation of RNA-seq data.

Gene Names Primers for the Oxidative Stress Response Gene Expression Assay
Forward Sequence (

Preparation of Samples for Transcriptomics
Untreated and 30-min H 2 O 2 (3 mM) treated cultures (0.4 at OD 600 ) of S. Typhimurium were centrifuged at 4000 rpm for 5 min. Supernatants were discard and the cell pellets were washed two times followed by RNA extraction using the RNeasy Mini kit (Qiagen Inc., Valencia, CA, USA) according to the manufacturer's instructions.

Global Transcriptomic Analysis
Sample quality preparation was done using capillary electrophoresis (Agilent Bio Analyzer 2100, Santa Clara, CA, USA) as previously described [17]. Only samples that had RNA integrity number 8 Antioxidants 2020, 9, 849 4 of 20 or greater were considered for further analysis. Illumina sequencing libraries were prepared using Illumina's TruSeq Stranded Total RNA Library Prep Human/Mouse/Rat Sample Preparation Kit (Cat. # 20020597) at the University of Minnesota Genomics Center (Saint Paul, MN, USA). For rRNA depletion, 500 ng of total amount of RNA was used in combination with Ribozero capture probes. After this step, the mRNA was fragmented followed by reverse transcription into complementary DNA (cDNA). The resulting cDNA fragments were coded with indexed adaptors followed by amplification using 15 PCR cycles. Again capillary electrophoresis was used to validate library size distribution, while the library quantification was carried out using fluorimetry. Libraries normalization was carried out followed by their hybridization to a single read flow cell. Once hybridization completed, the flow cell was loaded on the HiSeq 2500 and sequenced using Illumina's SBS chemistry. The primary analysis and index de-multiplexing were performed as previously described [17]. Read mapping was carried out using publicly available genome of S. Enteritidis P125109 strain. Quantification of gene expression was completed using Feature Counts. Significantly differentially expressed genes were identified using the edgeR feature in CLCGWB (Qiagen, Valencia, CA, USA) based on a minimum 2 times absolute fold change difference and p < 0.05 false discovery rate (FDR).

Validation of RNA-Seq Data by Real-Time PCR
Cultures of S. Enteritidis were prepared as described above and after 30 min of oxidative treatment samples were taken and processed. Total RNA was isolated as described above. Quantitative PCR assay was carried out as described above. Genes rtcR and yjjA were selected as the internal reference genes to normalize the expression of the tested genes. Primers were designed to target expression of the following genes: dinP, dnaJ, eutC, mdlA, osmY, pagO, spaR, malK, and nrdD (Table 1).

Experimental Replication and Gene Ontology Analysis
The data of all experiments represent the average of three biological replications. The cell viability during the oxidative treatments were analyzed by CoStat version 6.4 software (Co-Hort Software, Monterey, CA, USA) using the homogeneity of linear regression slopes approach. The gene ontology (GO) analysis was carried out using the Database for Annotation, Visualization and Integrated Discovery (DAVID) database [21] as well as the National Center for Biotechnology Information (NCBI).

RN-Seq Accession Numbers
RNA sequencing data for all three biological replications and two different treatments were deposited in the NCBI under accession number GSE 155479.

Effect ofIincreasing H 2 O 2 Concentrations on the Viability of S. Enteritidis
To evaluate the ability of S. Enteritidis to survive exposure to oxidative stress, we performed oxidative stress killing assays with increasing concentrations of H 2 O 2 over a short time. The concentration of 1 mM of H 2 O 2 caused a death rate of 0.37 log 10 CFU for the first 15 min of exposure ( Figure 1 and Figure S1). After this initial lethal period, the culture of S. Enteritidis exposed to the same concentration of H 2 O 2 showed a slight decline in the number of viable cells, resulting in only 0.1 log 10 CFU of reduction during the entire experiment ( Figure 1). The exposure of S. Enteritidis to 1 mM of H 2 O 2 resulted in two distinct survival phases, an initial and brief lethal phase followed by a long bacteriostatic phase (Figure 1 and Figure S1). Similarly to 1 mM, the concentration of 2 mM of H 2 O 2 caused an initial lethal phase resulting in 0.53 log 10 CFU decline ( Figure 1). However, the survival pattern associated with the 2 mM H 2 O 2 treatment was bactericidal not bacteriostatic ( Figure 1). After an initial lethal period for the first 15 min of H 2 O 2 exposure, the 2 mM H 2 O 2 treatment continued to cause cell death, albeit at a slower rate compared to that of the initial 15-min period. This bactericidal phase was characterized by repetitive fluctuations of fast die-off periods (e.g., 0.23, 0.24 and 0.22 log 10 Antioxidants 2020, 9, 849 5 of 20 CFU declines at the 30-min, 60-min, and 90-min measurements, respectively) and slight growth periods (e.g., 0.06 log 10 CFU increase at the 45-min measurement and 0.03 log 10 CFU increase at the 75-min measurement) (Figure 1). The exposure of S. Enteritidis to the greatest concentration of H 2 O 2 , 4 mM, caused again an initial lethal phase, characterized by a fast die-off rate, 0.72 log 10 CFU decline over the first 15 min of treatment ( Figure 1). Interestingly, this initial fast die-off period was followed by moderate die-off periods (e.g., 0.18 and 0.17 log 10 CFU decline) at the next two measurements (Figure 1). After this moderate die-off period, the culture of S. Enteritidis entered progressively fast mortality rates resulting in a 0.38 log 10 CFU decline at the 60-min measurement and 1.19 log 10 CFU decline at the 75-min measurement (Figure 1). At the last, 90-min, measurement, the viable cell count could not be detected at 10 3 CFU per 1 mL, indicating a continuation of this progressive and fast mortality rate. The oxidative treatments with 1 mM, 2 mM, and 4 mM of H 2 O 2 caused significantly higher (p < 0.01) mortality rates during the exponential growth phase of S. Enteritidis compared with no treatment of this pathogen under the same growth conditions. The average mortality rates of S. Enteritidis for every 15 min of H 2 O 2 treatment were 0.08 log 10 CFU decline for 1 mM, followed by 0.188 log 10 CFU decline for 2 mM treatment and 0.468 log 10 CFU decline for 4 mM. Although all three H 2 O 2 treatments caused significantly higher mortality rates compared to that of no treatment, the survival patterns of H 2 O 2 treated S. Enteritidis were different ( Figure 1). Antioxidants 2020, 9, x FOR PEER REVIEW  5 of 19 75-min measurement) ( Figure 1). The exposure of S. Enteritidis to the greatest concentration of H2O2, 4 mM, caused again an initial lethal phase, characterized by a fast die-off rate, 0.72 log10 CFU decline over the first 15 min of treatment ( Figure 1). Interestingly, this initial fast die-off period was followed by moderate die-off periods (e.g., 0.18 and 0.17 log10 CFU decline) at the next two measurements ( Figure 1). After this moderate die-off period, the culture of S. Enteritidis entered progressively fast mortality rates resulting in a 0.38 log10 CFU decline at the 60-min measurement and 1.19 log10 CFU decline at the 75-min measurement (Figure 1). At the last, 90-min, measurement, the viable cell count could not be detected at 10 3 CFU per 1 mL, indicating a continuation of this progressive and fast mortality rate. The oxidative treatments with 1 mM, 2 mM, and 4 mM of H2O2 caused significantly higher (p < 0.01) mortality rates during the exponential growth phase of S. Enteritidis compared with no treatment of this pathogen under the same growth conditions. The average mortality rates of S. Enteritidis for every 15 min of H2O2 treatment were 0.08 log10 CFU decline for 1 mM, followed by 0.188 log10 CFU decline for 2 mM treatment and 0.468 log10 CFU decline for 4 mM. Although all three H2O2 treatments caused significantly higher mortality rates compared to that of no treatment, the survival patterns of H2O2 treated S. Enteritidis were different ( Figure 1).

Expression Levels of the Genes Associated with Oxidative Stress and Anabolic Processes during Incremental Increase of H2O2 Concentration
To determine the relationship between a death rate of S. Enteritidis and expression of genes that are associated with the most important oxidative stress adaptive processes, we carried out a qRT-PCR assay. The gene expression assay was carried out after 30 min of H2O2 treatment when Salmonella cells have already been adapted to the oxidative stressor ( Figure 1). The targeted genes were selected from the following stress response and anabolism associated processes, iron homeostasis (iroN, sitA), alternative sigma factors (rpoS, rpoH), multiple stress responses (ycfR), DNA protection (dps), antioxidants (trxC), DNA replication (nrdM), uptake of small solutes (ompF) and anabolic process (pocR).
No significant changes in expression of rpoS were observed during 1 mM and 2 mM H2O2 treatment, whereas 4 mM H2O2 treatment caused a significant (2.5-fold) downregulation of the rpoS gene ( Figure 2). Similarly to the rpoS gene, the oxidative treatments of S. Enteritidis caused minor alterations in the expression of the gene that encodes another stress response sigma factor, RpoH

Expression Levels of the Genes Associated with Oxidative Stress and Anabolic Processes during Incremental Increase of H 2 O 2 Concentration
To determine the relationship between a death rate of S. Enteritidis and expression of genes that are associated with the most important oxidative stress adaptive processes, we carried out a qRT-PCR assay. The gene expression assay was carried out after 30 min of H 2 O 2 treatment when Salmonella cells have already been adapted to the oxidative stressor ( Figure 1). The targeted genes were selected from the following stress response and anabolism associated processes, iron homeostasis (iroN, sitA), alternative sigma factors (rpoS, rpoH), multiple stress responses (ycfR), DNA protection (dps), antioxidants (trxC), DNA replication (nrdM), uptake of small solutes (ompF) and anabolic process (pocR).
No significant changes in expression of rpoS were observed during 1 mM and 2 mM H 2 O 2 treatment, whereas 4 mM H 2 O 2 treatment caused a significant (2.5-fold) downregulation of the rpoS gene ( Figure 2). Similarly to the rpoS gene, the oxidative treatments of S. Enteritidis caused minor alterations in the expression of the gene that encodes another stress response sigma factor, RpoH ( Figure 2). The 1 mM and 4 mM treatments caused 2-fold and 2.9-fold upregulation of the rpoH gene, while 2 mM treatment caused no significant (1.4-fold) upregulation of the same gene.
Antioxidants 2020, 9, x FOR PEER REVIEW 6 of 19 ( Figure 2). The 1 mM and 4 mM treatments caused 2-fold and 2.9-fold upregulation of the rpoH gene, while 2 mM treatment caused no significant (1.4-fold) upregulation of the same gene. In sharp contrast to the genes that encode stress response sigma factors, the genes associated with iron homeostasis, in particular genes that encode the iron receptors, IroN (out membrane receptor) and SitA (periplasmic receptor) showed the greatest gene expression alterations ( Figure 2). Most notably, the expression of iroN exhibited a profound upregulation with the 1 mM H2O2 treatment (532-fold) and with 2 mM H2O2 treatment, the upregulation of this gene was extraordinarily high (75664-fold) compared to that of the H2O2 untreated control. Interestingly, with the greatest concentration of oxidative stressor, 4 mM, the upregulation of the iroN gene was lower (292-fold) than that with 1 mM (532 fold) and 2 mM (75664 fold) treatment ( Figure 2). Similarly, other genes involved in the iron homeostasis, sitA, showed large upregulation upon the oxidative stress treatment with 1 mM (298-fold), 2mM (316-fold) and 4 mM (675-fold) H2O2 treatment. Although both genes encode for iron receptors, their expression (stress response) patterns were different. The upregulation of the sitA gene was proportional to incremental increases of H2O2 concentration, whereas the upregulation of the iroN gene did not follow the level of H2O2 concentration ( Figure 2).
The ycfR gene, which encodes a multiple stress response protein YcfR, was upregulated throughout the H2O2 treatments. The upregulation of the ycfR gene ranged from 283-fold and 273fold to 103-fold in the Salmonella cultures exposed to 1 mM, 2 mM, and 4 mM H2O2, respectively ( Figure 2). Similarly to the iroN gene, the greatest concentration of H2O2, 4 mM, caused the lowest upregulation of the ycfR gene, further indicating to some extent the similar patterns of the responses of these two oxidative stress response associated genes. However, the ycfR gene did not show an extremely high upregulation during the 2 mM H2O2 treatment, as the iroN gene showed ( Figure 2). Another stress response gene, dps, had an unusual expression profile. During treatment with 1 mM H2O2 this gene was upregulated 44-fold, then during the 2 mM treatment the dps gene was downregulated 4-fold, and again upregulated 8.9-fold with 4 mM H2O2 treatment ( Figure 2). This unusual expression profile of the dps gene, which encodes a protein that protects DNA from reactive In sharp contrast to the genes that encode stress response sigma factors, the genes associated with iron homeostasis, in particular genes that encode the iron receptors, IroN (out membrane receptor) and SitA (periplasmic receptor) showed the greatest gene expression alterations ( Figure 2). Most notably, the expression of iroN exhibited a profound upregulation with the 1 mM H 2 O 2 treatment (532-fold) and with 2 mM H 2 O 2 treatment, the upregulation of this gene was extraordinarily high (75664-fold) compared to that of the H 2 O 2 untreated control. Interestingly, with the greatest concentration of oxidative stressor, 4 mM, the upregulation of the iroN gene was lower (292-fold) than that with 1 mM (532 fold) and 2 mM (75664 fold) treatment ( Figure 2). Similarly, other genes involved in the iron homeostasis, sitA, showed large upregulation upon the oxidative stress treatment with 1 mM (298-fold), 2mM (316-fold) and 4 mM (675-fold) H 2 O 2 treatment. Although both genes encode for iron receptors, their expression (stress response) patterns were different. The upregulation of the sitA gene was proportional to incremental increases of H 2 O 2 concentration, whereas the upregulation of the iroN gene did not follow the level of H 2 O 2 concentration ( Figure 2).
The ycfR gene, which encodes a multiple stress response protein YcfR, was upregulated throughout the H 2 O 2 treatments. The upregulation of the ycfR gene ranged from 283-fold and 273-fold to 103-fold in the Salmonella cultures exposed to 1 mM, 2 mM, and 4 mM H 2 O 2 , respectively ( Figure 2). Similarly to the iroN gene, the greatest concentration of H 2 O 2 , 4 mM, caused the lowest upregulation of the ycfR gene, further indicating to some extent the similar patterns of the responses of these two oxidative stress response associated genes. However, the ycfR gene did not show an extremely high upregulation during the 2 mM H 2 O 2 treatment, as the iroN gene showed ( Figure 2). Another stress response gene, dps, had an unusual expression profile. During treatment with 1 mM H 2 O 2 this gene was upregulated 44-fold, then during the 2 mM treatment the dps gene was downregulated 4-fold, and again upregulated 8.9-fold with 4 mM H 2 O 2 treatment ( Figure 2). This unusual expression profile of the dps gene, which encodes a protein that protects DNA from reactive oxygen species (ROS), indicates a complex interaction of the oxidative stress response associated genes.
The genes associated with DNA replication and oxidoreductase processes, nrdM and trxC, had similar genes expressions profiles ( Figure 2). Both genes exhibited 35-to 70-fold upregulation during the oxidative treatments ( Figure 2). The ompF gene, which encodes for a major nonspecific OmpF porin, and the pocR gene, which is associated with anabolic processes, exhibited downregulation during the oxidative treatments ( Figure 2). Interestingly, the downregulation gene expression patterns for both the ompF and pocR genes were the same. The greatest gene alterations (e.g., downregulation) were observed during 2 mM H 2 O 2 treatment, whereas the 4 mM H 2 O 2 treatment caused the lowest level of the gene expression alteration for both genes (Figure 2).

Global Transcriptome Response of S. Enteritidis to Oxidative Stress.
After the determination of the stress response patterns of a limited number of genes during treatment with different concentrations of H 2 O 2 , we carried out the global transcriptome analyses using a single concentration of H 2 O 2 . In total, 2051 genes were significantly differently expressed (SDE) in S. Enteritidis during the 3 mM H 2 O 2 treatment compared to the reference transcriptome of the same S. Enteritidis strain with no H 2 O 2 treatment. Of the 2051 SDE genes, 1111 were induced and 941 genes were repressed, showing at least a 2-fold alteration in gene expression and high reproducibility (false discovery rate < 0.5) across all three biological replications (Tables S1 and S2).

Induction of Transcriptome-Molecular Response of S. Enteritidis to Oxidative Stress
The most altered change in biological processes of oxidatively stressed S. Enteritidis cells was the induction of genes encoding Fe-S cluster biogenesis (Tables S1 and S2). Operons encoding sulfur utilization factor (SUF), sufABCSD and ynhAG, were upregulated from 173-to 903-fold and from 23-to 114-fold, respectively ( Figure 3). The SUF operons were profoundly more induced compared to ones encoding iron-sulfur cluster (ISC), nifJUS, yfhFP and hscBA ( Figure 3). Another group of highly induced genes were associated with the iron ion homeostasis. The operon sitABCD, iron/manganese ABC transporter, showed the greatest induction among operons/genes associated with this biological function ( Figure 3). Besides this Fe/Mn transporter, a group of operons/genes encoding a wide range of iron transporters showed the greatest inductions, including iroN, ferric salmochelin siderophore receptor; fepBD, ferric enterobactin ABC transporter; fhuABCD, ferrichrome outer membrane transporter; ydiE, hemin transporter; exbBD, iron siderophore bacteriocin transporter; fhuE, ferric coprogen and ferric-rhodotorulic acid receptor as well as fhuF, receptor for ferrioxamine B ( Figure 3). Also, several operons and a gene responsible for the biosynthesis of salmochelin (iroBCE), enterobactin (ybdABZ, entABCDEF) and hem (hemH) were greatly induced during oxidative stress ( Figure 3).
The significant induction showed a group of operons/genes encoding proteins that govern the cell-redox homeostasis. Among the cell-redox encoding genes, the nrdH gene which encodes a classical antioxidant, glutaredoxin-like protein NrdH, showed the greatest induction, resulting in 144.3-fold upregulation compared to the reference strain ( Figure 4). The trxC gene that encodes an effective cytoplasmic disulfide-reducing protein, thioredoxin 2, showed 76-fold induction ( Figure 4). The grxAB operon, encoding the antioxidants glutaredoxin 1 and 2, showed a moderate induction compared to that of the nrdH gene. A major part of the cell-redox regulon was associated with the genes encoding proteins associated with the electron transport chain (Figure 4). The narIJHKG operon, involved in nitrate assimilation and anaerobic electron transport, exhibited the greatest induction among this group of genes, followed by fdoHGI, dehydrogenases that transfer electrons from the periplasmic to the cytoplasmic heme; sdhABCD, cytochrome b556; cyoABCDE, cytochrome bo3 oxidase complex and sthA, which converts NADPH to NADH (Figure 4).   compared to that of the nrdH gene. A major part of the cell-redox regulon was associated with the genes encoding proteins associated with the electron transport chain (Figure 4). The narIJHKG operon, involved in nitrate assimilation and anaerobic electron transport, exhibited the greatest induction among this group of genes, followed by fdoHGI, dehydrogenases that transfer electrons from the periplasmic to the cytoplasmic heme; sdhABCD, cytochrome b556; cyoABCDE, cytochrome bo3 oxidase complex and sthA, which converts NADPH to NADH (Figure 4).  The most numerous group of altered operons/genes in the transcriptome of oxidatively stressed S. Enteritidis cells was associated with various stress response processes as a distinct biological function indicated by the GO analysis. The greatest alteration in the entire transcriptome of S. Enteritidis under H2O2 stress was related to the ycfR gene. This gene, encoding the multiple stress resistance outer membrane protein YcfR, was induced 913-fold compared to that of the reference strain ( Figure 4). The great induction exhibited the dps gene, which encodes the DNA-binding protein Dps that protects DNA. Besides dps, another group of genes encoding DNA repair proteins was significantly induced, including the DNA damage inducible yebG gene and the recAN, polB, dinPIG, and deoBCD genes (Figure 4). Similarly to genes encoding DNA repair proteins, a group of genes encoding molecular chaperones and proteases was significantly induced. Among this group of genes, The most numerous group of altered operons/genes in the transcriptome of oxidatively stressed S. Enteritidis cells was associated with various stress response processes as a distinct biological function indicated by the GO analysis. The greatest alteration in the entire transcriptome of S. Enteritidis under H 2 O 2 stress was related to the ycfR gene. This gene, encoding the multiple stress resistance outer membrane protein YcfR, was induced 913-fold compared to that of the reference strain ( Figure 4). The great induction exhibited the dps gene, which encodes the DNA-binding protein Dps that protects DNA. Besides dps, another group of genes encoding DNA repair proteins was significantly induced, including the DNA damage inducible yebG gene and the recAN, polB, dinPIG, and deoBCD genes (Figure 4). Similarly to genes encoding DNA repair proteins, a group of genes encoding molecular chaperones and proteases was significantly induced. Among this group of genes, the ibpBA gene, encoding small heat shock proteins, showed the greatest induction, followed by molecular chaperones dnaKJ, yhhW, and clpAB, and protease pgtE (Figure 4).
The oxidative stress caused the induction of numerous operons/genes located on Salmonella pathogenicity islands (SPIs). The significant induction was observed for the ssrAB operon that positively controls the expression of the SPI-2 genes ( Figure 5)  . Indeed, a significant induction of the ssaMCBDEVJKHILQN operon encoding the type III secretion system (T3SS) apparatus of the SPI-2 as well as operons sifAB and sseAbJBbFIE encoding the SPI-2 effector proteins ( Figure 5) was observed. Also, spaQSROP (translocon), iagB, sinHR, and sopE2 (effectors) as well as sicP (effector chaperone) located on the SPI-1 were significantly induced ( Figure 5). Not only genes encoding the T3SS apparatus and effector proteins responsible for the invasion of host cells and intracellular survival, but also genes encoding various cell adhesions were induced. This group of genes included yncC, biofilm transcriptional regulator; stbABCD, fimHYWDF, sthAB and SEN4247, fimbrial operons/genes; safBCD and SEN1978, pilin operon/gene as well as csgABF, curli operon ( Figure 5).

Repression of Transcriptome-Molecular Response of S. Enteritidis to Oxidative Stress
There is a distinct segregation based on the biological functions between the induced and repressed transcriptome of oxidatively stressed S. Enteritidis cells. While the induced transcriptome showed a great association with the iron-sulfur cluster assembly, iron ion homeostasis, cell redox homeostasis, a variety of stress responses, pathogenesis and sessile lifestyle (cell adhesion), the repressed transcriptome was mainly associated with carbohydrate metabolic processes, ribosomal biogenesis, anaerobic respiration and cell motility ( Figure 6). Indeed, a significant induction of the ssaMCBDEVJKHILQN operon encoding the type III secretion system (T3SS) apparatus of the SPI-2 as well as operons sifAB and sseAbJBbFIE encoding the SPI-2 effector proteins ( Figure 5) was observed. Also, spaQSROP (translocon), iagB, sinHR, and sopE2 (effectors) as well as sicP (effector chaperone) located on the SPI-1 were significantly induced ( Figure 5). Not only genes encoding the T3SS apparatus and effector proteins responsible for the invasion of host cells and intracellular survival, but also genes encoding various cell adhesions were induced. This group of genes included yncC, biofilm transcriptional regulator; stbABCD, fimHYWDF, sthAB and SEN4247, fimbrial operons/genes; safBCD and SEN1978, pilin operon/gene as well as csgABF, curli operon ( Figure 5).

Repression of Transcriptome-Molecular Response of S. Enteritidis to Oxidative Stress
There is a distinct segregation based on the biological functions between the induced and repressed transcriptome of oxidatively stressed S. Enteritidis cells. While the induced transcriptome showed a great association with the iron-sulfur cluster assembly, iron ion homeostasis, cell redox homeostasis, a variety of stress responses, pathogenesis and sessile lifestyle (cell adhesion), the repressed transcriptome was mainly associated with carbohydrate metabolic processes, ribosomal biogenesis, anaerobic respiration and cell motility ( Figure 6). . Indeed, a significant induction of the ssaMCBDEVJKHILQN operon encoding the type III secretion system (T3SS) apparatus of the SPI-2 as well as operons sifAB and sseAbJBbFIE encoding the SPI-2 effector proteins ( Figure 5) was observed. Also, spaQSROP (translocon), iagB, sinHR, and sopE2 (effectors) as well as sicP (effector chaperone) located on the SPI-1 were significantly induced ( Figure 5). Not only genes encoding the T3SS apparatus and effector proteins responsible for the invasion of host cells and intracellular survival, but also genes encoding various cell adhesions were induced. This group of genes included yncC, biofilm transcriptional regulator; stbABCD, fimHYWDF, sthAB and SEN4247, fimbrial operons/genes; safBCD and SEN1978, pilin operon/gene as well as csgABF, curli operon ( Figure 5).

Repression of Transcriptome-Molecular Response of S. Enteritidis to Oxidative Stress
There is a distinct segregation based on the biological functions between the induced and repressed transcriptome of oxidatively stressed S. Enteritidis cells. While the induced transcriptome showed a great association with the iron-sulfur cluster assembly, iron ion homeostasis, cell redox homeostasis, a variety of stress responses, pathogenesis and sessile lifestyle (cell adhesion), the repressed transcriptome was mainly associated with carbohydrate metabolic processes, ribosomal biogenesis, anaerobic respiration and cell motility ( Figure 6). that showed the greatest alterations compared to their counterparts in the control sample. Each gene showed false discovery rate (FDR) p < 0.05 and high reproducibility across the biological replications.

Validation of RNA-seq Data by Real-Time PCR
To validate the RNA-seq data, the expressions of nine randomly chosen genes were tested by quantitative real-time PCR (qRT-PCR) under the same experimental conditions. The qRT-PCR showed the same pattern of expressions, including seven upregulated genes (dinP, dnaJ, eutC, mdlA, osmY, pagO, and spaR) and two downregulated genes (malK and nrdD) (Figure 8). It can be observed that generally, the expression values measured by qRT-PCR were greater compared to those measured by RNA-seq analysis. The difference in expression values between these two techniques is consistent with the fact that qRT-PCR generally gives higher expression values compared to RNAseq [17,22]. Carbohydrate metabolism was the biological function most commonly affected within the repressed transcriptome. This group of operons/genes, encoding a wide range of proteins, was involved in catabolic pathways, including gudDP (D-glucarate), garDLKR (galactarate), eutSPQT (ethanolamine), nanAEKT (sialic acid), melABR (melibiose), srlAR (sorbitol), kdgTK (D-glucuronate), idnDT, and gntRT (D-gluconate) (Figure 7). Besides this group of operons/genes associated with catabolism, there was repression of operons/genes encoding sugar uptake and anabolism such as malFTKMG (maltose), ydeWZYV (carbohydrate permease), yadBEI (a major carbohydrate active transport system), mglABC (galactose/methyl galactoside) and citCDEFD2E2 (citric operon involved in anabolic processes of the tricarboxylic acid cycle) (Figure 7).

Validation of RNA-seq Data by Real-Time PCR
To validate the RNA-seq data, the expressions of nine randomly chosen genes were tested by quantitative real-time PCR (qRT-PCR) under the same experimental conditions. The qRT-PCR showed the same pattern of expressions, including seven upregulated genes (dinP, dnaJ, eutC, mdlA, osmY, pagO, and spaR) and two downregulated genes (malK and nrdD) (Figure 8). It can be observed that generally, the expression values measured by qRT-PCR were greater compared to those measured by RNA-seq analysis. The difference in expression values between these two techniques is consistent with the fact that qRT-PCR generally gives higher expression values compared to RNA-seq [17,22]. Antioxidants 2020, 9, x FOR PEER REVIEW 15 of 19 . Figure 8. Validation of RNA-seq data by qRT-PCR analysis. Data represent fold changes in the expression of nine randomly selected genes between the wild type strain without treatment and the same strain treated with 3 mM H2O2. Genes differently expressed between the wild type strain with no treatment and the wild type with H2O2 treatment represent the mean value of three biological replications.

Discussion
Salmonella Enteritidis, one of the most important non-typhoidal Salmonella serovars from the public health and veterinary medicine perspectives [23][24][25][26], has developed a robust anti-oxidative response not only to survive endogenous oxidative stress [27] but also to survive exogenous stress imposed by a host [28] or environment [29]. In other words, for successful host invasion, intracellular proliferation and/or environmental survival, possession of an efficient oxidative stress response is one of the key physiological characteristics of this pathogen. In this study, it was shown that incrementally increased concentrations of oxidant agent caused various patterns of gene expressions. While inductions of genes encoding antioxidant (thioredoxin 2) and DNA replication regulator (NrdM) were constant regardless of H2O2 concentrations, induction of a gene encoding iron/manganese transporter, SitA, was proportional with H2O2 concentrations. In contrast to the pattern of sitA induction, the ycfR gene, encoding multiple stress-response protein, exhibited disproportionate induction regarding H2O2 concentration. The greatest concentration of H2O2 resulted in the lowest ycfR induction. Besides these three patterns of gene expressions, the most common gene expression alteration during this assay was associated with an unpredictable pattern, where the middle concentration of H2O2 caused the greatest gene expression alterations, either induction (iroN) or repression (pocR, ompF and dps). Recently, Mitosch et al. [30], using a dualreported model, demonstrated the importance of timing of the gene expression during exposure of the microbial cell to an oxidative agent. Besides the importance of temporal gene expression response, this study points to a multifaceted effect of oxidative agent concentration on gene expression during oxidative stress, further indicating another dimension of cellular adaptation complexity to oxidative stress.
The comparative transcriptomic analyses revealed that the expression of 45% of the S. Enteritidis genome was altered at least two-fold upon exposure to H2O2. This finding showed that oxidative stress has a much greater impact on the physiology of this pathogen than previously anticipated [18,31,32]. Out of numerous biological functions altered by oxidative stress, the SUF mechanism of Fe-S cluster formation exhibited the greatest change. The Fe-S clusters are cofactors of proteins and they are involved in various critical metabolic processes including DNA replication, regulation, and repair, substrate uptake, respiration, and RNA modification [33]. In general, there are two Fe-S cluster formation systems, the ISC system and the SUF system [33]. In this study, the SUF system showed a significantly greater induction than the ISC system. This finding correlates with previous studies that

Discussion
Salmonella Enteritidis, one of the most important non-typhoidal Salmonella serovars from the public health and veterinary medicine perspectives [23][24][25][26], has developed a robust anti-oxidative response not only to survive endogenous oxidative stress [27] but also to survive exogenous stress imposed by a host [28] or environment [29]. In other words, for successful host invasion, intracellular proliferation and/or environmental survival, possession of an efficient oxidative stress response is one of the key physiological characteristics of this pathogen. In this study, it was shown that incrementally increased concentrations of oxidant agent caused various patterns of gene expressions. While inductions of genes encoding antioxidant (thioredoxin 2) and DNA replication regulator (NrdM) were constant regardless of H 2 O 2 concentrations, induction of a gene encoding iron/manganese transporter, SitA, was proportional with H 2 O 2 concentrations. In contrast to the pattern of sitA induction, the ycfR gene, encoding multiple stress-response protein, exhibited disproportionate induction regarding H 2 O 2 concentration. The greatest concentration of H 2 O 2 resulted in the lowest ycfR induction. Besides these three patterns of gene expressions, the most common gene expression alteration during this assay was associated with an unpredictable pattern, where the middle concentration of H 2 O 2 caused the greatest gene expression alterations, either induction (iroN) or repression (pocR, ompF and dps). Recently, Mitosch et al. [30], using a dual-reported model, demonstrated the importance of timing of the gene expression during exposure of the microbial cell to an oxidative agent. Besides the importance of temporal gene expression response, this study points to a multifaceted effect of oxidative agent concentration on gene expression during oxidative stress, further indicating another dimension of cellular adaptation complexity to oxidative stress.
The comparative transcriptomic analyses revealed that the expression of 45% of the S. Enteritidis genome was altered at least two-fold upon exposure to H 2 O 2 . This finding showed that oxidative stress has a much greater impact on the physiology of this pathogen than previously anticipated [18,31,32]. Out of numerous biological functions altered by oxidative stress, the SUF mechanism of Fe-S cluster formation exhibited the greatest change. The Fe-S clusters are cofactors of proteins and they are involved in various critical metabolic processes including DNA replication, regulation, and repair, substrate uptake, respiration, and RNA modification [33]. In general, there are two Fe-S cluster formation systems, the ISC system and the SUF system [33]. In this study, the SUF system showed a significantly greater induction than the ISC system. This finding correlates with previous studies that pointed out that the ISC is a housekeeping system [34], while the SUF was shown to be an oxidative stress or Fe-limited inducible system [35]. Indeed, the transcriptomic analyses showed the significant induction of numerous Fe acquisition mechanisms, including an iron/manganese transporter, ferrichrome, hemin, ferrioxamine, ferric coprogen and ferric-rhodotorulic acid uptake systems, then biosynthesis and uptake of salmochelin, enterobactin, as well as hydroxamate siderophores, clearly indicating a state of Fe-limitation in oxidatively stressed cells. These data showed the activation of the various Fe-acquisition systems in S. Enteritidis cells during oxidative stress, further suggesting their critical importance in cellular adaptation to oxidative stress.
Hydroxyl free radical, a product of the Fenton reaction, represents the most damaging compound during H 2 O 2 treatment [36]. This highly reactive oxygen species interacts with numerous biomolecules further causing multiple effects on the treated prokaryotic cells [37]. The comparative transcriptomics analyses showed multifaceted responses of oxidatively treated cells including induction of not only the OxyR (katG, ahpCF) and the SoxRS (soxR, soxS, sodA, sodC), two oxidative stress regulons, but also stress responses involved in DNA and protein repairs, osmoprotection, multidrug efflux, phosphate and magnesium starvation. During the oxidative stress response, this human pathogen induced a robust DNA repair system, which included base repair (deoBCD), single base mismatch repair (dinIG), early double-strand break repair (yebG), multiple double-strand breaks repair (recAN) and DNA replication (polB, dinP). Also, the dps gene which encodes a multifunctional Dps protein was greatly induced. Dps non-specifically binds DNA, forming a highly stable Dps-DNA nucleoprotein complex which further protects DNA from the deleterious effects of ROS [38]. Similarly to DNA, protein repair systems were induced. The ibpAB genes, encoding small heat-shock proteins IbpA and IbpB, showed the greatest induction among various protein repair systems. Induction of genes encoding molecular chaperones (dnaKJ, clpB) and proteases (clpA, pgtE) indicates a tendency of proteins to aggregate or to undergo a process of denaturation during oxidative stress. Multifaceted responses of oxidatively treated cells was also reflected by induction of key genes associated with osmotic stress (osmYC, envFR), manganese starvation (sitABCD, pagCDOK, mgtBC), phosphate starvation (phoEH, yjbA, phnTSUR), multidrug efflux (mdlAB, yhcQP) and outer membrane modification (wcaEFH, yncCJ, yjbAEHJ). Adaptation of S. Enteritidis to oxidative stress was not only confined to the induced transcriptome but it also included the repressed transcriptome. The significant repression of the major small (rpsUTPLGSOCFHNJQBR) and large (rplUKSJNXVBCDAEPLWYM) units of ribosomal encoding genes indicates that oxidatively stressed cells activated a stringent response. The main purpose of a stringent response is saving cellular energy and precursors required for ribosomal synthesis. If cells continue to synthesize ribosomes, the majority of these organelles would end up being inactive due to the disruption of central metabolic pathways (i.e., lack of amino acids) and this would be an unnecessary and costly expenditure of resources [37].
Besides the alterations of a large pool of genes encoding various central metabolic and stress response pathways, the comparative transcriptome analyses revealed a clear division, during the H 2 O 2 treatment, in the gene expressions associated with the genes encoding for motility (planktonic) and the adhesion genes (sessile) forms of life. The exposure of S. Enteritidis to oxidative stress had an extensive and significant positive effect on the expression of adhesion encoding genes, including fimbrial stbABCD, fimHYWDF, sthB, SEN4247; curli csgABF, and pilin safBCD, SEN1978 (type IV pilin), yibEHG and yncC operons and genes. At the same time, it negatively affected expressions of the major flagellar operons flgHECFIGJDBMN and fliJMRLNKFQPOIHEBG clearly indicating a switch from a planktonic to a sessile (i.e., biofilm) life forms of oxidatively treated cells. Recently, it was shown that a transition from planktonic to sessile life form in S. Enteritidis is followed by repression of the major flagellar operons and induction of numerous fimbrial, curli, and pili operons [22]. This transcriptional change is in agreement with the "swim-or-stick" theory, which postulates that motility and biofilm development are mutually-exclusive processes [39]. Our transcriptomic data indicate that oxidatively stressed cells transition from a planktonic to a biofilm form of life. Another important feature of oxidatively stressed cells was an extensive induction of salmonellae-essential virulence genes located on SPI 1, SPI 2, and SPI 5. Induction of core virulence genes involved genes encoding not only T3SS apparatus of the SPI 2 and translocon of SPI 1 but also numerous effectors (sifAB, sseAbjbBFIE, iagB, sinHR, sopE2, sicP) essential for intracellular survival of this pathogen [40]. Fu and colleagues [18] studying the oxidative stress response of S. Typhimurium found that this organism repressed the synthesis of core virulence proteins encoded on SPI 1. Discrepancy between their and our results can be most likely explained by different experimental setups. While Fu and colleagues [18] exposed the cultures of S. Typhimurium to H 2 O 2 at their mid-exponential growth phases (0.5 at OD 600 ) and harvested them at their early stationary growth phase (0.9 at OD 600 ). We started H 2 O 2 treatment at the mid-exponential growth phase (0.4 at OD 600 ) and harvested the cultures at their still exponential growth phase (~0.5 at OD 600 ). Taken together, our experimental approach ensured that all transcriptional changes will be consequences of oxidative stress, while the approach employed by Fu et al., [18] reflects not only oxidative stress response but additional stress responses associated with the stationary growth phase of this organism [41,42].

Conclusions
This study employing a combination of incremental oxidative stress assays and the next generation sequencing approach revealed additional levels of the oxidative stress response complexity in non-typhoidal Salmonella. It was shown that the oxidative stress response is intricately linked with not only Fe-S cluster formation, iron homeostasis, DNA repair, and cell redox homeostasis but also with heat, osmotic, stringent, manganese and phosphate starvation stress responses as well as multidrug efflux and decreased passive influx of small hydrophilic molecules. In addition to these multifaceted cellular adaptation processes, it was shown that S. Enteritidis during oxidative stress simultaneously repressed the key motility encoding genes and massively induced the essential adhesion and salmonellae core-virulence encoding genes, crucial for the biofilm formation and host cell invasion. Based on these data, we hypothesize that the oxidative stress response of non-typhoidal Salmonella is closely associated with intracellular survival and that oxidative stress may act as a stimulus for eukaryotic cell invasion and formation of Salmonella-containing vacuoles. It has been already found that intracellular biofilm formation plays an important step in virulence of uropathogenic Escherichia coli, another facultative intracellular pathogen [43], suggesting that a similar approach can be explored by non-typhoidal Salmonella.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-3921/9/9/849/s1, Figure S1. The effect of different H 2 O 2 concentrations on the survivability of S. Enteritidis over 90 min. Error bars correspond to the standard deviations. Table S1. List of induced genes during the 3 mM H 2 O 2 treatment of the S. Enteritidis ATCC 13076 strain compared to the control. Table S2